Don: >I think this change is mandatory. We need it for SIMD operations.<
Why? Why the compiler can't optimize things and perform SIMD operations with the fixed-sized array semantics of D1? (I ask this for LDC too, that's mostly D1 still). Bye, bearophile
